Article Social Identity Reduces Depression by
Fostering Positive Attributions Social Psychological and
Personality Science
2015, Vol. 6(1) 65-74
ª The Author(s) 2014
Reprints and permission:
sagepub.com/journalsPermissions.nav
DOI: 10.1177/1948550614543309
spps.sagepub.com Tegan Cruwys1, Erica I. South1, Katharine H. Greenaway1,
and S. Alexander Haslam1 Abstract
Social identities are generally associated with better health and in particular lower levels of depression. However, there has been
limited investigation of why social identities protect against depression. The current research suggests that social identities reduce
depression in part because they attenuate the depressive attribution style (internal, stable, and global; e.g., ‘‘I failed because I’m
stupid’’). These relationships are first investigated in a survey (Study 1, N ¼ 139) and then followed up in an experiment that
manipulates social identity salience (Study 2, N ¼ 88). In both cases, people with stronger social identities were less likely to
attribute negative events to internal, stable, or global causes and subsequently reported lower levels of depression. These studies
thus indicate that social identities can protect and enhance mental health by facilitating positive interpretations of stress and
failure. Implications for clinical theory and practice are discussed.
Keywords
depression, social identity, multiple group membership, attribution, failure, mental health
We cannot live only for ourselves.
A thousand fibers connect us with our fellow men.
—Herman Melville, Moby Dick Humans have an innate need for social connections that are vital for
health and happiness in life (Baumeister & Leary, 1995; Cohen &
Wills, 1985). When this need is not met—when the ‘‘thousand
fibers’’ in Melville’s (1851) quote are reduced to a few or
none—people are at risk of reduced well-being and even mental illness. In fact, depression—the leading cause of disability worldwide
(World Health Organization, 2012)—commonly arises when a person lacks social connections (Cacioppo, Hawkley, & Thisted,
2010; Cacioppo, Hughes, Waite, Hawkley, & Thisted, 2006).
Although extensive prior work has documented the fact that
social connectedness and social identities are critical to mental
health and reduced rates of depression, it remains unclear why
this is the case (Cruwys, Haslam, Dingle, Haslam, & Jetten,
2014; Jetten, Haslam, Haslam, Dingle, & Jones, 2014). We
address this research gap in the present work. What exactly do
social identities do, psychologically, that makes them so protective for mental health? In answering this question, we propose a
novel mechanism through which social identities can protect people against depression, that is, reduced depressive attributions. Social Identity and Depression
A growing body of literature demonstrates that social identities
are a key psychological resource that is protective for health generally (Haslam, Jetten, Postmes, & Haslam, 2009; Jetten,
Haslam, & Haslam, 2012) and against depression in particular
(Cruwys et al., 2013; Cruwys, Haslam, Dingle, Haslam, et al,
2014; Cruwys, Haslam, Dingle, Jetten, et al., 2014). Social
identity refers to that part of the self-concept that reflects the
internalization of group memberships (Tajfel & Turner,
1979). This means that when the self is defined by a given
social identity (e.g., ‘‘us Catholics,’’ ‘‘us psychologists,’’ and
‘‘us Australians’’), we see other members of that group not as
‘‘other’’ but as part of who and what we are. Moreover, when
groups are internalized in this way, they provide us with a sense
of belonging, purpose, and direction and therefore have the
capacity to enrich our lives—in particular, by providing a basis
for bonding, support, companionship, and security (Haslam
et al., 2009).
The psychological resources that social identity provides
in turn have positive consequences for health and well-being
(Jetten et al., 2012). Indeed, growing evidence suggests that
social identities have an important role in preventing and resolving depression. For example, in a longitudinal study with a
nationally representative sample of over 5,000 older adults, 1 University of Queensland, St Lucia, Queensland, Australia
Erica I. South and Tegan Cruwys are joint first authors.
Corresponding Author:
Tegan Cruwys, School of Psychology, The University of Queensland, St Lucia,
Queensland, 4072, Australia.
Email: t.cruwys@uq.edu.au 66
Cruwys et al. (2013) found that possessing multiple group
memberships protected against the development of depression,
improved the likelihood of recovering from depression, and
prevented depression relapse. Indeed, evidence suggested that
each new social group an individual joined reduced their risk
of relapse 4 years later by 24%. It is also worth noting that several studies provide evidence that it is primarily social isolation
that leads to depression, with only limited evidence for the
opposite causal pathway (Cacioppo et al., 2010; Iyer, Jetten,
Tsivrikos, Postmes, & Haslam, 2009).
One clue that social identification is at the heart of this process is that groups need to be psychologically important to an
individual in order to reduce depression symptoms (Cruwys,
Haslam, Dingle, Haslam, et al, 2014; Wakefield, Bickley, &
Sani, 2013). For instance, high social identification with the
army was associated with lower depression among soldiers
(Sani, Herrera, Wakefield, Boroch, & Gulyas, 2012), and students who report high identification with an educational institution also report lower levels of depression (Bizumic, Reynolds,
Turner, Bromhead, & Subasic, 2009; Brook, Garcia, & Fleming,
2008; Iyer et al., 2009). Similar effects were found in an intervention study that encouraged people with depression to join
social groups: reductions in depression symptoms were most
marked for those individuals with high (rather than low) levels of social identification (Cruwys, Haslam, Dingle, Jetten,
et al., 2014).
Accumulating evidence thus points to the role of social identity as an active antidote to depression—both that having more
social identities and that identifying more strongly with any
particular social group protect against the condition. Yet while
there is strong evidence for such effects, it nonetheless remains
unclear exactly why joining social groups has this positive protective effect. Accordingly, there is clearly a need to explore
mechanisms through which social identity might reduce
depression.
Speaking to this issue, previous research has suggested
that the relationship between social identity and depression
might be mediated by social support (Haslam, O’Brien, Jetten,
Vormedal, & Penna, 2005; Jetten et al., 2014; Sani, 2012),
since a sense of shared social identity has been shown to be a
basis for both the provision of help and its positive construal
(e.g., Haslam, Reicher, & Levine, 2012; Levine, Prosser,
Evans, & Reicher, 2005). However, given that just thinking
about one’s social identities has the capacity to improve
well-being and resilience (e.g., Jones & Jetten, 2011), it would
seem likely that candidate mediators will also be psychological
and not (just) material. In this regard too, it is important to note
that both theoretical and empirical works speak to the capacity
for social identity to fundamentally restructure cognition (e.g.,
Turner, Oakes, Haslam, & McGarty, 1994). For example, studies have found that when individuals define themselves in
terms of shared social identity (rather than as separate individuals; Turner, 1982), they are less paranoid, more empathic, and
more cognitively engaged (Branscombe & Miron, 2004;
Haslam et al., 2014; Reicher & Haslam, 2006). Other research
has shown that social identity has a profound impact on the way Social Psychological and Personality Science 6(1)
people interpret and explain the social world—as reflected in
stereotypic attributions (Oakes, Haslam, & Turner, 1994;
Oakes, Turner, & Haslam, 1991) and in cognitive processing
more generally (e.g., McGarty, Yzerbyt, & Spears, 2002). A
key question, then, is whether such cognitive restructuring
might have implications for depression. Social Identity and Attribution Style
One hallmark of depression is a negative attribution style
when generating causal explanations for events (Peterson &
Seligman, 1984; Sweeney, Anderson, & Bailey, 1986; Weiner,
Nierenberg, & Goldstein, 1976). Specifically, individuals with
a depressive attribution style are more likely to attribute negative events to causes that are internal, stable across time, and
influence many areas of their life (Abramson, Seligman, &
Teasdale, 1978). Positive events are explained in the opposite
way—as externally caused, transient, and situation specific.
This pattern of thinking leads people to blame themselves for
failure while denying credit for success (Sweeney et al., 1986).
A large body of evidence suggests that this depressive attribution style is causally related to depression (Chan, 2012;
Peterson & Seligman, 1984; Sweeney et al., 1986). Therefore,
if a person’s depressive attribution style is altered so as to
become more positive (i.e., external, transient, and specific),
then depression symptoms should be reduced (Seligman
et al., 1988). In line with other work showing that social identities serve to structure attributions (e.g., Oakes et al., 1991), it
is therefore relevant to ask whether social identity might reduce
depression by attenuating depressive attribution style.
There are several reasons for hypothesizing that this might
be the case. First, a large body of research has shown that causal explanations can be altered by social factors. In particular,
attribution style has been found to develop in response to social
influences, such as children modeling their parents’ explanatory style (Haines, Metalsky, Cardamone, & Joiner, 1999; Lau,
Belli, Gregory, Napolitano, & Eley, 2012; Seligman et al.,
1984). Moreover, once a depressive attribution style develops,
it can still be modified, even in adults (Seligman et al., 1988).
Speaking to this possibility, Klein, Fencil-Morse, and Seligman
(1976) had depressed and nondepressed students complete
unsolvable problems. Attributions for failure were manipulated
by telling participants that most people succeeded on the task
(inducing an internal attribution for personal failure) or that
most people failed on the task (inducing an external attribution
for personal failure). After experiencing failure, depressed individuals tend to perform poorly on subsequent tasks. However,
inducing an external attribution of failure reduced the number
of depressed individuals who adopted a depressive attribution
style. More importantly, this external attribution eliminated
subsequent poor performance among depressed individuals.
In other words, simply being made aware that personal behavior was similar to that of other group members produced more
positive attributions, reducing symptoms of depression and
associated poor performance. Cruwys et al.
One way in which social identity could change such attributions is by shifting attentional focus. Depressed individuals
have previously been found to have a self-focusing style such
that they tend to focus internally—that is on their personal
selves—following failure (Greenberg, Pyszczynski, Burling,
& Tibbs, 1992). This focus on personal shortcomings leads to
internal attributions for failure, causing depression (Greenberg
et al., 1992; Romens, MacCoon, Abramson, & Pollak, 2011). In
contrast, thinking about one’s social identity shifts attention
away from the self as an individual (Hogg & Williams, 2000;
Turner & Onorato, 1999) and toward (generally positive) group
memberships (Turner, 1982). Thus, when social identity is
made salient, one’s shortcomings as an individual may be less
salient, reducing the chance of internal attributions for failure.
In addition to reducing internal attributions for failure,
social identity may also facilitate internal attributions for success. This is important because, as a corollary to the patterns
discussed earlier, depressed individuals have been found to
lack the common (personal) self-serving attribution style in
which credit is taken for personal success and blame is denied
for personal failure (Greenberg et al., 1992; Seidel et al., 2012).
Yet while depressed individuals typically fail to exhibit this
form of self-serving bias, there is some evidence they
still engage in a group-serving attribution bias (Dietz-Uhler
& Murrell, 1998). For example, Schlenker and Britt (1996)
found that depressed individuals had a depressive attribution
style when explaining their own and strangers’ experiences but
that when attributing the same events for their close friends,
they made positive attributions—apportioning credit for their
successes and minimizing blame for their failures. This suggests that depressed individuals are capable of making positive
attributions and that social identities may provide a cognitive
platform for them to do so. The Present Research
The goal of the present research was to explore the interrelationships between social identity, depressive attributions, and
depression. In line with the above-mentioned reasoning, we
tested the following four core hypotheses:
Hypothesis 1: social identity would be associated with
reduced depression,
Hypothesis 2: social identity would be associated with positive attribution styles,
Hypothesis 3: positive attribution styles would be associated
with reduced depression, and
Hypothesis 4: a decrease in depressive attributions will
mediate the protective effect of social identity on depression.
Importantly, while Hypothesis 1 is supported by previous
social identity research (e.g., Cruwys, Haslam, Dingle, Haslam,
et al., 2014; Cruwys, Haslam, Dingle, Jetten, et al., 2014;
Reicher & Haslam, 2006; Sani et al., 2012) and Hypothesis 3
by a large body of prior work on depression (e.g., Peterson &
Seligman, 1984; Sweeney et al., 1986; Weiner et al., 1976), 67
to our knowledge this is the first research to propose and test
Hypothesis 2 and Hypothesis 4 and to explore the cognitive
processes that mediate the relationship between social identity
and mental health more generally.
For this purpose, we conducted two studies. The first was a
survey study designed to explore the various relationships postulated in the above-mentioned hypotheses. The second was an
experiment in which we manipulated social identity salience
with a view to establishing its causal impact on attribution style
and depression. Study 1
Study 1 surveyed final-year university students in the process of
completing a major research thesis in psychology. This sample is
at high risk of depression (Murphy, Gray, Sterling, Reeves, &
DuCette, 2009; Stallman, 2010) and therefore particularly appropriate for investigating our hypotheses. Furthermore, questionnaires were completed during the week that students were
submitting their research thesis, and for this reason the sample
was expected to be under considerable acute stress. Social identity was operationalized in this study as multiple group memberships, in line with previous research suggesting that these act as a
psychological resource that protects against depression (Cruwys
et al., 2013; Iyer et al., 2009). Method
Participants and Design. An online questionnaire was administered to 139 final-year psychology students (Mage ¼ 23.93,
SD ¼ 5.07; 115 female) from four universities. The questionnaire included measures of multiple group memberships,
depressive attribution style, and depression symptoms. A range
of other variables related to personality and mental health were
also measured but do not relate to our hypothesis and will not
be discussed further.
Materials
Multiple group memberships. The Exeter Identity Transition
Scale (Haslam et al., 2008) was used to gauge participants’
involvement in multiple groups. The scale comprised 7 items
(e.g., ‘‘I am active in lots of different groups’’). Participants
rated their agreement with each item on a scale from 1 (not
at all) to 7 (completely), a ¼ .95.
Depressive attribution style. The Depressive Attributions
Questionnaire (DAQ; Kleim, Gonzalo, & Ehlers, 2011) was
used to measure depressive attribution style. This scale comprised 16 items across four related attribution areas (internal,
stable, global, and perceived helplessness) such as ‘‘when bad
things happen, I think it is my fault’’ on a 4-point scale from 0
(not at all) to 4 (very strongly), a ¼ .91.
Depression symptoms. The Centre for Epidemiological Studies Depression Scale (Radloff, 1977) was used to assess current
levels of depression. Participants responded to 20 statements
that asked how often they had experienced symptoms of 68 Social Psychological and Personality Science 6(1) depression over the last week. Response options varied from
1 (rarely or none of the time) to 4 (most or all of the time),
a ¼ .93. Results
Descriptives Statistics and Analytic Strategy. The average level of
depression was high (M ¼ 18.43; SD ¼ 12.31), exceeding the
diagnostic cutoff of 16. Average depressive attribution style
was also higher than previously reported norms (M ¼ 31.01;
SD ¼ 8.58, compared to M ¼ 20.52 in three studies by Kleim
et al., 2011). Students reported a moderate level of involvement
in groups (M ¼ 4.02; SD ¼ 1.60). This is similar to previous
research with students transitioning to university (Iyer et al.,
2009). Regression analyses were conducted to assess Hypotheses 1, 2, and 3.
Hypothesis 1: Multiple group memberships protect against
depression. Results supported Hypothesis 1, with multiple
group memberships significantly predicting depression
scores, R2 ¼ .07, F(1, 137) ¼ 11.66, b ¼ .28, p ¼ .001. Participants who reported having more group memberships
tended to have lower levels of depression symptoms. Among
those with an above average number of group memberships,
38% were above the cutoff for depression; however, among
those with a below average number of groups, 65% met criteria for depression.
Hypothesis 2: Multiple group memberships promote positive
attributions. Analyses also supported Hypothesis 2, R2 ¼ .05,
F(1, 137) ¼ 6.47, b ¼ .21, p ¼ .012. Participants with multiple
group memberships had a more positive attribution style. Those
with higher levels of multiple group memberships (1 SD above
the mean) scored 4 points lower on the DAQ than those with
fewer group memberships (1 SD below the mean).
Hypotheses 3 and 4: Positive attributions mediate the protective
effect of group memberships. Regression analysis also supported
Hypothesis 3. When entered in a second block following multiple group memberships, depressive attribution style significantly predicted depression symptoms, R2D ¼ .24, F(1, 137)
¼ 48.23, b ¼ .50, p < .001. Among respondents who had a
more positive attribution style (i.e., below the sample mean),
32% exceeded the cutoff score for depression, whereas among
those with a more negative attribution style (above the mean),
70% met this cutoff. The impact of multiple group memberships also became less pronounced when depressive attribution
style was entered in the model, b ¼ .17, p ¼ .018.
In order to test Hypothesis 4, a mediation analysis with
10,000 bootstrap samples was conducted (Hayes, 2013, model
4). Multiple group memberships were included as the predictor,
with depression symptoms as the outcome variable. Depressive
attribution style was entered as the mediator. The indirect
effect (IE) of multiple group memberships was significant
(IE ¼ .11, standard error [SE] ¼ .04, 95% confidence interval, CI [.20, .02]). In line with Hypothesis 4, a decrease Depressive Attribution
Style
.50**
-.21* Multiple Group
Memberships Depression
Symptoms
-.28** (-.17*) Figure 1. Depressive attribution style partially mediates the
relationship between multiple group memberships and depression
symptoms. All numbers are standardized coefficients.
Note. N ¼ 139. *p < .05. **p < .001. in depressive attribution style partially mediated the effect of
multiple group memberships on lower levels of depression
symptoms. The model is displayed in Figure 1.
We also tested the most plausible alternative mediation
model, in which depressive attributions predicted reduced multiple group memberships via increased depression symptoms.
This model did not provide a good fit to the data, as it explained
only 8% of the variance (compared to 32% of the variance in
the hypothesized model) and attribution style was not a significant direct predictor of multiple group memberships (b ¼
.09, p ¼ .377). However, the IE of attribution style on multiple group memberships was significant (IE ¼ .13, SE ¼ .06,
95% CI [.25, .02]). Discussion
As hypothesized, participants with more group memberships
had lower levels of depression (Hypothesis 1) and were less
likely to make depressive attributions (Hypothesis 2). Moreover, depressive attributions were associated with higher
depressive symptoms (Hypothesis 3) and partially mediated the
relationship between multiple group membership and depression (Hypothesis 4). This study provides preliminary support
for the proposition that social identity can reduce depression
by attenuating depressive attribution style. In a sample of
highly stressed students, the availability of multiple social
identities protected against depression by encouraging a more
positive attribution style.
Although an alternative model predicting lower group memberships through depression was also significant, this model
was less powerful than our hypothesized model. Of course,
with correlational data it is impossible to definitively determine
causal ordering of variables. To address this limitation, we
therefore conducted an experiment in which we manipulated
social identity salience. Study 2
Study 2 investigated whether an experimental manipulation of
social identity salience could reduce depressive attributions
and negative mood following failure on a task. This design
would allow us to more confidently infer the causal role of
social identity in shaping depressive attributions. Negative Cruwys et al.
mood was used as an analogue of depression because it was not
feasible to alter an individual’s depression symptoms in an
experimental context (for similar logic, see Klein et al.,
1976; Spielberger, Ritterband, Reheiser, & Brunner, 2003).
Social identities were made salient by asking participants in
three independent conditions to reflect on no groups (control),
one group, or three groups that they belong to. Rather than
investigating attributions during a period of high stress (as in
Study 1), all participants in Study 2 experienced failure on a
problem-solving task—a context in which, theoretically,
depressive attributions are most problematic.
The study tested the same four hypotheses as in the previous
study. More specifically, we anticipated that participants in the
social identity conditions (one group or three groups) would
have reduced negative mood (Hypothesis 1) and make more
positive attributions (Hypothesis 2) than those in the control
condition and that positive attributions would reduce negative
mood (Hypothesis 3) and mediate the relationship between
social identity salience and negative mood (Hypothesis 4). Method
Participants and Design. Participants were 88 undergraduate psychology students (Mage ¼ 19.72, SD ¼ 3.86; 55 females) who
received partial course credit for their participation. Participants were randomly assigned to one of the three conditions
(none vs. one vs. three groups) and identity salience was
manipulated between subjects. After the manipulation, all participants completed a problem-solving task and received failure
feedback before completing the dependent measures.
Materials and Measures
Identity manipulation. In the control condition, participants
did not complete a written reflection task. In the two other conditions, participants first read a short paragraph explaining
what constitutes a social identity (adapted from Haslam, Oakes,
Reynolds, & Turner, 1999). Participants in the one-group condition then listed one group they belonged to and wrote about
‘‘why this group is an important part of who you are.’’ Participants in the three-group condition completed the same exercise, listing three groups (Jones & Jetten, 2011). Most
participants took less than 5 min to complete the task.
Failure paradigm. The failure paradigm was presented to participants as a four-question problem-solving task (following
Carver & Scheier, 1982; Klein et al., 1976; Welch & Huston,
1982). The task involved four unsolvable questions developed in
accordance with previous attribution research (e.g., Mikulincer,
1989; Koole, Smeets, vanKnippenberg, & Dijksterhuis,
1999). The test was collected from participants after 5 minutes.

Attachments:

• 1506065894-19485506145433091.pdf